def download_file(cls, uri, fobj):
"""
Given a URI, download the file to the given file-like object.
:param str uri: The URI of a file to download.
:param file fobj: A file-like object to download the file to.
:rtype: file
:returns: A file handle to the downloaded file.
"""
# Breaks the URI into usable componenents.
values = get_values_from_media_uri(uri)
conn = cls._get_aws_s3_connection(values['username'],
values['password'])
bucket = conn.get_bucket(values['host'])
key = bucket.get_key(values['path'])
logger.debug("S3Backend.download_file(): " \
"Downloading: %s" % uri)
dlhandler = ResumableDownloadHandler(num_retries=10)
try:
dlhandler.get_file(key, fobj, None)
except AttributeError:
# Raised by ResumableDownloadHandler in boto when the given S3
# key can't be found.
message = "The specified input file cannot be found."
raise InfileNotFoundException(message)
logger.debug("S3Backend.download_file(): " \
"Download of %s completed." % uri)
return fobj
def test_failed_download_with_persistent_tracker(self):
"""
Tests that failed resumable download leaves a correct tracker file
"""
harness = CallbackTestHarness()
tmpdir = self._MakeTempDir()
tracker_file_name = self.make_tracker_file(tmpdir)
dst_fp = self.make_dst_fp(tmpdir)
res_download_handler = ResumableDownloadHandler(
tracker_file_name=tracker_file_name, num_retries=0)
small_src_key_as_string, small_src_key = self.make_small_key()
try:
small_src_key.get_contents_to_file(
dst_fp, cb=harness.call,
res_download_handler=res_download_handler)
self.fail('Did not get expected ResumableDownloadException')
except ResumableDownloadException, e:
# We'll get a ResumableDownloadException at this point because
# of CallbackTestHarness (above). Check that the tracker file was
# created correctly.
self.assertEqual(e.disposition,
ResumableTransferDisposition.ABORT_CUR_PROCESS)
self.assertTrue(os.path.exists(tracker_file_name))
f = open(tracker_file_name)
etag_line = f.readline()
self.assertEquals(etag_line.rstrip('\n'), small_src_key.etag.strip('"\''))
def test_download_with_file_content_change_during_download(self):
"""
Tests resumable download on an object where the file content changes
without changing length while download in progress
"""
harnass = CallbackTestHarnass(
fail_after_n_bytes=self.larger_src_key_size / 2,
num_times_to_fail=2)
# Set up first process' ResumableDownloadHandler not to do any
# retries (initial download request will establish expected size to
# download server).
res_download_handler = ResumableDownloadHandler(
tracker_file_name=self.tracker_file_name, num_retries=0)
dst_filename = self.dst_fp.name
try:
self.larger_src_key.get_contents_to_file(
self.dst_fp,
cb=harnass.call,
res_download_handler=res_download_handler)
self.fail('Did not get expected ResumableDownloadException')
except ResumableDownloadException, e:
# First abort (from harnass-forced failure) should be
# ABORT_CUR_PROCESS.
self.assertEqual(e.disposition,
ResumableTransferDisposition.ABORT_CUR_PROCESS)
# Ensure a tracker file survived.
self.assertTrue(os.path.exists(self.tracker_file_name))
def test_multiple_in_process_failures_then_succeed_with_tracker_file(self):
"""
Tests resumable download that fails completely in one process,
then when restarted completes, using a tracker file
"""
# Set up test harness that causes more failures than a single
# ResumableDownloadHandler instance will handle, writing enough data
# before the first failure that some of it survives that process run.
harness = CallbackTestHarness(
fail_after_n_bytes=LARGE_KEY_SIZE/2, num_times_to_fail=2)
larger_src_key_as_string = os.urandom(LARGE_KEY_SIZE)
larger_src_key = self._MakeKey(data=larger_src_key_as_string)
tmpdir = self._MakeTempDir()
tracker_file_name = self.make_tracker_file(tmpdir)
dst_fp = self.make_dst_fp(tmpdir)
res_download_handler = ResumableDownloadHandler(
tracker_file_name=tracker_file_name, num_retries=0)
try:
larger_src_key.get_contents_to_file(
dst_fp, cb=harness.call,
res_download_handler=res_download_handler)
self.fail('Did not get expected ResumableDownloadException')
except ResumableDownloadException, e:
self.assertEqual(e.disposition,
ResumableTransferDisposition.ABORT_CUR_PROCESS)
# Ensure a tracker file survived.
self.assertTrue(os.path.exists(tracker_file_name))
def _pull_key(self, key, fname):
Logger.debug("Pulling key '{}' from bucket '{}' to file '{}'".format(
key.name, key.bucket.name, fname))
self._makedirs(fname)
tmp_file = self.tmp_file(fname)
name = os.path.basename(fname)
if self._cmp_checksum(key, fname):
Logger.debug('File "{}" matches with "{}".'.format(
fname, key.name))
return fname
Logger.debug('Downloading cache file from S3 "{}/{}" to "{}"'.format(
key.bucket.name, key.name, fname))
res_h = ResumableDownloadHandler(
tracker_file_name=self._download_tracker(tmp_file), num_retries=10)
try:
key.get_contents_to_filename(tmp_file,
cb=create_cb(name),
res_download_handler=res_h)
except Exception as exc:
Logger.error('Failed to download "{}": {}'.format(key.name, exc))
return None
os.rename(tmp_file, fname)
progress.finish_target(name)
Logger.debug('Downloading completed')
return fname
def test_download_with_inital_partial_download_before_failure(self):
"""
Tests resumable download that successfully downloads some content
before it fails, then restarts and completes
"""
# Set up harness to fail download after several hundred KB so download
# server will have saved something before we retry.
harness = CallbackTestHarness(
fail_after_n_bytes=LARGE_KEY_SIZE/2)
larger_src_key_as_string = os.urandom(LARGE_KEY_SIZE)
larger_src_key = self._MakeKey(data=larger_src_key_as_string)
res_download_handler = ResumableDownloadHandler(num_retries=1)
dst_fp = self.make_dst_fp()
larger_src_key.get_contents_to_file(
dst_fp, cb=harness.call,
res_download_handler=res_download_handler)
# Ensure downloaded object has correct content.
self.assertEqual(LARGE_KEY_SIZE,
get_cur_file_size(dst_fp))
self.assertEqual(larger_src_key_as_string,
larger_src_key.get_contents_as_string())
# Ensure some of the file was downloaded both before and after failure.
self.assertTrue(
len(harness.transferred_seq_before_first_failure) > 1 and
len(harness.transferred_seq_after_first_failure) > 1)
def test_failed_download_with_persistent_tracker(self):
"""
Tests that failed resumable download leaves a correct tracker file
"""
harnass = CallbackTestHarnass()
res_download_handler = ResumableDownloadHandler(
tracker_file_name=self.tracker_file_name, num_retries=0)
try:
self.small_src_key.get_contents_to_file(
self.dst_fp,
cb=harnass.call,
res_download_handler=res_download_handler)
self.fail('Did not get expected ResumableDownloadException')
except ResumableDownloadException, e:
# We'll get a ResumableDownloadException at this point because
# of CallbackTestHarnass (above). Check that the tracker file was
# created correctly.
self.assertEqual(e.disposition,
ResumableTransferDisposition.ABORT_CUR_PROCESS)
self.assertTrue(os.path.exists(self.tracker_file_name))
f = open(self.tracker_file_name)
etag_line = f.readline()
m = re.search(ResumableDownloadHandler.ETAG_REGEX, etag_line)
f.close()
self.assertTrue(m)
def _import(self, bucket_name, key_name, fname, data_item):
bucket = self._get_bucket_aws(bucket_name)
tmp_file = self.tmp_file(fname)
name = os.path.basename(fname)
key = bucket.get_key(key_name)
if not key:
Logger.error(
'File "{}" does not exist in the cloud'.format(key_name))
return None
if self._cmp_checksum(key, fname):
Logger.debug('File "{}" matches with "{}".'.format(
fname, key_name))
return data_item
Logger.debug('Downloading cache file from S3 "{}/{}" to "{}"'.format(
bucket.name, key_name, fname))
res_h = ResumableDownloadHandler(
tracker_file_name=self._download_tracker(tmp_file), num_retries=10)
try:
key.get_contents_to_filename(tmp_file,
cb=create_cb(name),
res_download_handler=res_h)
os.rename(tmp_file, fname)
except Exception as exc:
Logger.error('Failed to download "{}": {}'.format(key_name, exc))
return None
progress.finish_target(name)
Logger.debug('Downloading completed')
return data_item
def test_download_with_inconsistent_etag_in_tracker(self):
"""
Tests resumable download with an inconsistent etag in tracker file
"""
tmp_dir = self._MakeTempDir()
dst_fp = self.make_dst_fp(tmp_dir)
small_src_key_as_string, small_src_key = self.make_small_key()
inconsistent_etag_tracker_file_name = os.path.join(tmp_dir,
'inconsistent_etag_tracker')
f = open(inconsistent_etag_tracker_file_name, 'w')
good_etag = small_src_key.etag.strip('"\'')
new_val_as_list = []
for c in reversed(good_etag):
new_val_as_list.append(c)
f.write('%s\n' % ''.join(new_val_as_list))
f.close()
res_download_handler = ResumableDownloadHandler(
tracker_file_name=inconsistent_etag_tracker_file_name)
# An error should be printed about the expired tracker, but then it
# should run the update successfully.
small_src_key.get_contents_to_file(
dst_fp, res_download_handler=res_download_handler)
self.assertEqual(SMALL_KEY_SIZE,
get_cur_file_size(dst_fp))
self.assertEqual(small_src_key_as_string,
small_src_key.get_contents_as_string())
def test_zero_length_object_download(self):
"""
Tests downloading a zero-length object (exercises boundary conditions).
"""
res_download_handler = ResumableDownloadHandler()
self.empty_src_key.get_contents_to_file(
self.dst_fp, res_download_handler=res_download_handler)
self.assertEqual(0, get_cur_file_size(self.dst_fp))
def test_zero_length_object_download(self):
"""
Tests downloading a zero-length object (exercises boundary conditions).
"""
res_download_handler = ResumableDownloadHandler()
dst_fp = self.make_dst_fp()
k = self._MakeKey()
k.get_contents_to_file(dst_fp,
res_download_handler=res_download_handler)
self.assertEqual(0, get_cur_file_size(dst_fp))
def test_download_without_persistent_tracker(self):
"""
Tests a single resumable download, with no tracker persistence
"""
res_download_handler = ResumableDownloadHandler()
self.small_src_key.get_contents_to_file(
self.dst_fp, res_download_handler=res_download_handler)
self.assertEqual(self.small_src_key_size,
get_cur_file_size(self.dst_fp))
self.assertEqual(self.small_src_key_as_string,
self.small_src_key.get_contents_as_string())
def test_download_without_persistent_tracker(self):
"""
Tests a single resumable download, with no tracker persistence
"""
res_download_handler = ResumableDownloadHandler()
dst_fp = self.make_dst_fp()
small_src_key_as_string, small_src_key = self.make_small_key()
small_src_key.get_contents_to_file(
dst_fp, res_download_handler=res_download_handler)
self.assertEqual(SMALL_KEY_SIZE, get_cur_file_size(dst_fp))
self.assertEqual(small_src_key_as_string,
small_src_key.get_contents_as_string())
def test_multiple_in_process_failures_then_succeed(self):
"""
Tests resumable download that fails twice in one process, then completes
"""
res_download_handler = ResumableDownloadHandler(num_retries=3)
self.small_src_key.get_contents_to_file(
self.dst_fp, res_download_handler=res_download_handler)
# Ensure downloaded object has correct content.
self.assertEqual(self.small_src_key_size,
get_cur_file_size(self.dst_fp))
self.assertEqual(self.small_src_key_as_string,
self.small_src_key.get_contents_as_string())
def download_file(self, filename, destination=None):
if not destination:
destination = filename
key = self.bucket.get_key(filename)
handler = ResumableDownloadHandler(tempfile.mktemp())
handler._save_tracker_info(key) # Ugly but necessary
self.rate_limiter.reset( self.rate_limit, -1 )
with file(destination, "a") as f:
handler.get_file(key, f, {}, cb=self.callbacks, num_cb=-1)
#validate
md5 = utils.md5_for_large_file(destination)
s3_md5 = key.metadata.get('nimbus-md5', None)
if s3_md5:
if s3_md5 != md5:
raise utils.Md5CheckError("md5 mismatch")
else:
if not '-' in key.etag:
s3_md5 = key.etag.strip('"\'')
if md5 != s3_md5:
raise utils.Md5CheckError("md5 mismatch")
else:
size = os.path.getsize(destination)
if size != key.size:
raise FileSizeError("error")
def test_download_with_unwritable_tracker_file(self):
"""
Tests resumable download with an unwritable tracker file
"""
# Make dir where tracker_file lives temporarily unwritable.
save_mod = os.stat(self.tmp_dir).st_mode
try:
os.chmod(self.tmp_dir, 0)
res_download_handler = ResumableDownloadHandler(
tracker_file_name=self.tracker_file_name)
except ResumableDownloadException, e:
self.assertEqual(e.disposition, ResumableTransferDisposition.ABORT)
self.assertNotEqual(
e.message.find('Couldn\'t write URI tracker file'), -1)
def test_non_retryable_exception_handling(self):
"""
Tests resumable download that fails with a non-retryable exception
"""
harnass = CallbackTestHarnass(
exception=OSError(errno.EACCES, 'Permission denied'))
res_download_handler = ResumableDownloadHandler(num_retries=1)
try:
self.small_src_key.get_contents_to_file(
self.dst_fp, cb=harnass.call,
res_download_handler=res_download_handler)
self.fail('Did not get expected OSError')
except OSError, e:
# Ensure the error was re-raised.
self.assertEqual(e.errno, 13)
def test_broken_pipe_recovery(self):
"""
Tests handling of a Broken Pipe (which interacts with an httplib bug)
"""
exception = IOError(errno.EPIPE, "Broken pipe")
harnass = CallbackTestHarnass(exception=exception)
res_download_handler = ResumableDownloadHandler(num_retries=1)
self.small_src_key.get_contents_to_file(
self.dst_fp, cb=harnass.call,
res_download_handler=res_download_handler)
# Ensure downloaded object has correct content.
self.assertEqual(self.small_src_key_size,
get_cur_file_size(self.dst_fp))
self.assertEqual(self.small_src_key_as_string,
self.small_src_key.get_contents_as_string())
def test_retryable_exception_recovery(self):
"""
Tests handling of a retryable exception
"""
# Test one of the RETRYABLE_EXCEPTIONS.
exception = ResumableDownloadHandler.RETRYABLE_EXCEPTIONS[0]
harnass = CallbackTestHarnass(exception=exception)
res_download_handler = ResumableDownloadHandler(num_retries=1)
self.small_src_key.get_contents_to_file(
self.dst_fp, cb=harnass.call,
res_download_handler=res_download_handler)
# Ensure downloaded object has correct content.
self.assertEqual(self.small_src_key_size,
get_cur_file_size(self.dst_fp))
self.assertEqual(self.small_src_key_as_string,
self.small_src_key.get_contents_as_string())
def _GetTransferHandlers(self, uri, key, file_size, upload):
"""
Selects upload/download and callback handlers.
We use a callback handler that shows a simple textual progress indicator
if file_size is above the configurable threshold.
We use a resumable transfer handler if file_size is >= the configurable
threshold and resumable transfers are supported by the given provider.
boto supports resumable downloads for all providers, but resumable
uploads are currently only supported by GS.
"""
config = boto.config
resumable_threshold = config.getint('GSUtil', 'resumable_threshold',
ONE_MB)
if file_size >= resumable_threshold:
cb = self._FileCopyCallbackHandler(upload).call
num_cb = int(file_size / ONE_MB)
resumable_tracker_dir = config.get(
'GSUtil', 'resumable_tracker_dir',
os.path.expanduser('~' + os.sep + '.gsutil'))
if not os.path.exists(resumable_tracker_dir):
os.makedirs(resumable_tracker_dir)
if upload:
# Encode the src bucket and key into the tracker file name.
res_tracker_file_name = (re.sub(
'[/\\\\]', '_', 'resumable_upload__%s__%s.url' %
(key.bucket.name, key.name)))
else:
# Encode the fully-qualified src file name into the tracker file name.
res_tracker_file_name = (re.sub(
'[/\\\\]', '_', 'resumable_download__%s.etag' %
(os.path.realpath(uri.object_name))))
tracker_file = '%s%s%s' % (resumable_tracker_dir, os.sep,
res_tracker_file_name)
if upload:
if uri.scheme == 'gs':
transfer_handler = ResumableUploadHandler(tracker_file)
else:
transfer_handler = None
else:
transfer_handler = ResumableDownloadHandler(tracker_file)
else:
transfer_handler = None
cb = None
num_cb = None
return (cb, num_cb, transfer_handler)
def get_file(self, obj_path, file_path, tracker_path):
key = self.bucket.get_key(obj_path)
if key is None:
return None
handler = ResumableDownloadHandler(tracker_path)
filename = os.path.basename(obj_path)
with make_progress_bar(filename, key.size) as pbar:
def callback(total_xfer, total_size):
pbar.update(total_xfer)
with open(file_path, 'ab') as file_:
key.get_contents_to_file(file_,
cb=callback,
num_cb=NUM_CB,
res_download_handler=handler)
def test_failed_and_restarted_download_with_persistent_tracker(self):
"""
Tests resumable download that fails once and then completes,
with tracker file
"""
harnass = CallbackTestHarnass()
res_download_handler = ResumableDownloadHandler(
tracker_file_name=self.tracker_file_name, num_retries=1)
self.small_src_key.get_contents_to_file(
self.dst_fp, cb=harnass.call,
res_download_handler=res_download_handler)
# Ensure downloaded object has correct content.
self.assertEqual(self.small_src_key_size,
get_cur_file_size(self.dst_fp))
self.assertEqual(self.small_src_key_as_string,
self.small_src_key.get_contents_as_string())
# Ensure tracker file deleted.
self.assertFalse(os.path.exists(self.tracker_file_name))
def test_download_with_unwritable_tracker_file(self):
"""
Tests resumable download with an unwritable tracker file
"""
# Make dir where tracker_file lives temporarily unwritable.
tmp_dir = self._MakeTempDir()
tracker_file_name = os.path.join(tmp_dir, 'tracker')
save_mod = os.stat(tmp_dir).st_mode
try:
os.chmod(tmp_dir, 0)
res_download_handler = ResumableDownloadHandler(
tracker_file_name=tracker_file_name)
except ResumableDownloadException as e:
self.assertEqual(e.disposition, ResumableTransferDisposition.ABORT)
self.assertNotEqual(
e.message.find('Couldn\'t write URI tracker file'), -1)
finally:
# Restore original protection of dir where tracker_file lives.
os.chmod(tmp_dir, save_mod)
def test_download_with_invalid_tracker_etag(self):
"""
Tests resumable download with a tracker file containing an invalid etag
"""
invalid_etag_tracker_file_name = (
'%s%sinvalid_etag_tracker' % (self.tmp_dir, os.sep))
f = open(invalid_etag_tracker_file_name, 'w')
f.write('3.14159\n')
f.close()
res_download_handler = ResumableDownloadHandler(
tracker_file_name=invalid_etag_tracker_file_name)
# An error should be printed about the invalid tracker, but then it
# should run the update successfully.
self.small_src_key.get_contents_to_file(
self.dst_fp, res_download_handler=res_download_handler)
self.assertEqual(self.small_src_key_size,
get_cur_file_size(self.dst_fp))
self.assertEqual(self.small_src_key_as_string,
self.small_src_key.get_contents_as_string())
def test_failed_and_restarted_download_with_persistent_tracker(self):
"""
Tests resumable download that fails once and then completes,
with tracker file
"""
harness = CallbackTestHarness()
tmpdir = self._MakeTempDir()
tracker_file_name = self.make_tracker_file(tmpdir)
dst_fp = self.make_dst_fp(tmpdir)
small_src_key_as_string, small_src_key = self.make_small_key()
res_download_handler = ResumableDownloadHandler(
tracker_file_name=tracker_file_name, num_retries=1)
small_src_key.get_contents_to_file(
dst_fp, cb=harness.call, res_download_handler=res_download_handler)
# Ensure downloaded object has correct content.
self.assertEqual(SMALL_KEY_SIZE, get_cur_file_size(dst_fp))
self.assertEqual(small_src_key_as_string,
small_src_key.get_contents_as_string())
# Ensure tracker file deleted.
self.assertFalse(os.path.exists(tracker_file_name))
def test_download_with_invalid_tracker_etag(self):
"""
Tests resumable download with a tracker file containing an invalid etag
"""
tmp_dir = self._MakeTempDir()
dst_fp = self.make_dst_fp(tmp_dir)
small_src_key_as_string, small_src_key = self.make_small_key()
invalid_etag_tracker_file_name = os.path.join(tmp_dir,
'invalid_etag_tracker')
f = open(invalid_etag_tracker_file_name, 'w')
f.write('3.14159\n')
f.close()
res_download_handler = ResumableDownloadHandler(
tracker_file_name=invalid_etag_tracker_file_name)
# An error should be printed about the invalid tracker, but then it
# should run the update successfully.
small_src_key.get_contents_to_file(
dst_fp, res_download_handler=res_download_handler)
self.assertEqual(SMALL_KEY_SIZE, get_cur_file_size(dst_fp))
self.assertEqual(small_src_key_as_string,
small_src_key.get_contents_as_string())
def test_multiple_in_process_failures_then_succeed_with_tracker_file(self):
"""
Tests resumable download that fails completely in one process,
then when restarted completes, using a tracker file
"""
# Set up test harnass that causes more failures than a single
# ResumableDownloadHandler instance will handle, writing enough data
# before the first failure that some of it survives that process run.
harnass = CallbackTestHarnass(
fail_after_n_bytes=self.larger_src_key_size / 2,
num_times_to_fail=2)
res_download_handler = ResumableDownloadHandler(
tracker_file_name=self.tracker_file_name, num_retries=0)
try:
self.larger_src_key.get_contents_to_file(
self.dst_fp,
cb=harnass.call,
res_download_handler=res_download_handler)
self.fail('Did not get expected ResumableDownloadException')
except ResumableDownloadException, e:
self.assertEqual(e.disposition, ResumableTransferDisposition.ABORT)
# Ensure a tracker file survived.
self.assertTrue(os.path.exists(self.tracker_file_name))
def test_download_with_object_size_change_between_starts(self):
"""
Tests resumable download on an object that changes sizes between inital
download start and restart
"""
harnass = CallbackTestHarnass(
fail_after_n_bytes=self.larger_src_key_size / 2,
num_times_to_fail=2)
# Set up first process' ResumableDownloadHandler not to do any
# retries (initial download request will establish expected size to
# download server).
res_download_handler = ResumableDownloadHandler(
tracker_file_name=self.tracker_file_name, num_retries=0)
try:
self.larger_src_key.get_contents_to_file(
self.dst_fp,
cb=harnass.call,
res_download_handler=res_download_handler)
self.fail('Did not get expected ResumableDownloadException')
except ResumableDownloadException, e:
self.assertEqual(e.disposition, ResumableTransferDisposition.ABORT)
# Ensure a tracker file survived.
self.assertTrue(os.path.exists(self.tracker_file_name))
def test_download_with_inconsistent_etag_in_tracker(self):
"""
Tests resumable download with an inconsistent etag in tracker file
"""
inconsistent_etag_tracker_file_name = (
'%s%sinconsistent_etag_tracker' % (self.tmp_dir, os.sep))
f = open(inconsistent_etag_tracker_file_name, 'w')
good_etag = self.small_src_key.etag.strip('"\'')
new_val_as_list = []
for c in reversed(good_etag):
new_val_as_list.append(c)
f.write('%s\n' % ''.join(new_val_as_list))
f.close()
res_download_handler = ResumableDownloadHandler(
tracker_file_name=inconsistent_etag_tracker_file_name)
# An error should be printed about the expired tracker, but then it
# should run the update successfully.
self.small_src_key.get_contents_to_file(
self.dst_fp, res_download_handler=res_download_handler)
self.assertEqual(self.small_src_key_size,
get_cur_file_size(self.dst_fp))
self.assertEqual(self.small_src_key_as_string,
self.small_src_key.get_contents_as_string())
